CCM.FF-K1 for Water Flo · CCM.FF-K1 for Water Flow Final Report 17 May 2007 4 of 75 second change to the schedule occurred when, due to delays in the commissioning of the new water

CCM.FF-K1 for Water Flow

Final Report

Jong S. Paik,1 Kwang-Bock Lee,

1 Peter Lau,

2 Rainer Engel,

3

Alejandro Loza,4 Yoshiya Terao,

5 Michael Reader-Harris

6

November, 2006

KRISS 1 Doryongdong, Yusong Gu, Daejon 305-340, Korea

[email protected]

1 KRISS, Korea

2 SP, Sweden

3 PTB, Germany

4 CENAM, Mexico

5 NMIJ, Japan

6 NEL, U.K.

CCM.FF-K1 for Water Flow Final Report

17 May 2007 2 of 75

Table of Content

1 Introduction ............................................................................................... 2

2 Organization .............................................................................................. 3

3 Transfer Standard Package (TSP) ............................................................. 4

4 Test Procedure........................................................................................... 6

5 Measurement Uncertainty ....................................................................... 11

6 Pilot Comparison Tests ........................................................................... 12

7 Key Comparison (KC) for Water Flow: CCM.FF-K1 ............................ 21

7.1. Description of the TSP for K1........................................................... 21

7.2. Technical Protocol............................................................................. 23

8 KC Results............................................................................................... 23

8.1. Reynolds Number Correction............................................................ 23

8.2. Results of Participating NMIs ........................................................... 23

8.2.1. Reproducibility variation....................................................... 23

8.2.2. K -factors............................................................................... 26

8.2.3. Installation effects.................................................................. 26

8.2.4. KCRVs and NMI-to-KCRV differences ............................... 31

8.2.5. NMI-to-NMI differences ....................................................... 44

9 Summary and Conclusions...................................................................... 74

10 Acknowledgements ................................................................................. 75

11 References ............................................................................................... 75

1 Introduction

In October 1999, the International Committee for Weights and Measures (CIPM), under

authority granted to it by the Convention of the Metre, signed into existence a Mutual

Recognition of National Measurement Standards and of Calibration and Measurement

Certificates Issued by National Metrology Institutes (CIPM MRA). The goal of the

CIPM MRA is to quantify the equivalence of the measurement standards maintained in

the world’s National Metrology Institutes (NMIs). This quantification is contained in an

electronic database, made available through the World Wide Web, which aids in the

reduction of measurement-based technical barriers to international trade (see KCDB).

The strategy of the CIPM MRA is to perform key comparisons (KCs) in which

participating NMIs can compare their performance through their calibration data [1]. As


17 May 2007 3 of 75

a result, the quantification of standards’ equivalence, assessed via the routine calibration

procedures used by the NMIs, assures buyers and sellers in different countries that the

measurements used in their commercial transactions are equivalent within their stated

uncertainties. It stands that the economic impact of the quality of the KCs is high and

for this reason, the resulting data and the conclusions drawn from it need to be of the

highest attainable quality.

This report describes the results obtained during a KC conducted in the area of water

flow metrology. This KC, known as CCM.FF-K1, is here described and its results are

used to give an indication of the equivalence of the water flow measurement standards

maintained by the participating NMIs. In the future, and in accord with the design of the

CIPM MRA, these results are to be linked, via additional KCs conducted by the

Regional Metrology Organizations (RMOs), so that the equivalence of water flow

standards extends to all interested NMIs worldwide. Additionally, CCM.FF-K1 is to be

repeated at intervals so that the equivalences are maintained over time.

2 Organization

The Korea Research Institute of Standards and Science (KRISS), served as pilot

laboratory for this KC and five (5) additional NMIs, representing 3 RMOs, participated:

for EUROMET: SP (SE), PTB (DE), and NEL (GB); for SIM: CENAM (MX); and for

APMP, NMIJ (JP) in addition to KRISS (KR).

The comparison was conducted by circulating a transfer standard package (TSP)

among the participants. As the KC progressed, the TSP traveled from one participant to

the next and each NMI was responsible for the customs clearance and delivery of the

TSP to the next NMI.

The proposed schedule of CCM.FF-K1 is shown in Table 1, however unexpected

circumstances led to a couple of changes. The first change came about when a box

containing the TSP’s flow meters and pipe pieces was found to be damaged upon arrival

to Japan. After visual inspection, the TSP was tested and the resulting data did not show

any indication of TSP deterioration. In spite of this, the pilot laboratory decided to

perform in-house testing of the TSP before sending it to NEL. The damaged box was

replaced by a new one provided by NMIJ prior to sending it to KRISS. This led to a

two-week delay and the TSP was shipped to NEL from KRISS on May 3, 2004. The


17 May 2007 4 of 75

second change to the schedule occurred when, due to delays in the commissioning of

the new water facility at NEL, the comparison was delayed until November, 2004.

3 Transfer Standard Package (TSP)

The aim of CCM.FF-K1 was to compare the national water flow standards as realized in

the participating NMIs and this demanded that the KC be conducted at the highest

metrological level currently possible. The participants claimed that they could perform

water flow measurements with uncertainties of %1.0± or better. In order to quantify

such measurement uncertainties, a stable high precision7 TSP and comprehensive KC

test procedure were needed. Specifically, the test procedure needed to generate

measurements which were statistically sufficient so attain a high confidence level of the

desired results.

Table 1. Proposed schedule for CCM.FF-K1.

RMO NMI Date Remark

APMP KRISS (KR) 16 Sept 03 –15 Oct 03 PILOT Lab

EUROMET SP (SE) 16 Oct 03 –30 Nov 03

EUROMET PTB (DE) 1 Dec 03 –15 Jan 04

SIM CENAM (MX) 16 Jan 04 –29 Feb 04 PIVOT Lab

APMP NMIJ (JP) 1 March 04 –15 Apr 04

EUROMET NEL (GB) 16 Apr 04 –31 May 04 PIVOT Lab

APMP KRISS (KR) 1 June 04 –30 June 04 PILOT Lab

Data Analysis & Report 1 July 04 –15 Aug 04

To that aim, a TSP comprised of two stable high-precision flow meters was designed so

that high confidence could be placed in the results of the KC while allowing the

participating NMIs to use their routine capabilities to calibrate the TSP. A third flow

meter was included in the package to monitor the flow profile entering the TSP. Figure

1 illustrates the transfer standard package (TSP), which consisted of 3 pipe units of

7 Accuracy of measurement: closeness of the agreement between the results of a measurement and the

true value of the measurand.

NOTES

1 “Accuracy” is a qualitative concept.

2 The term precision should not be used for “accuracy”.

Def. 3.5, International Vocabulary of Basic and General Terms in Metrology (VIM), ed. 1993.


17 May 2007 5 of 75

0.1016 m in nominal diameter (schedule 40, ANSI 150 lb raised face), 3 flow meters,

associated instrumentation and data acquisition system, with a total length of 3.36 m. In

this configuration, the 3 flow meters were: (1) ultrasonic flow meter (Changmin Tech,

UR-1000), (2) coriolis meter (Endress+Hauser, Promass 83), and (3) turbine meter

(Exact Flow, EFM64DR-W-C-T1-B-X0). The coriolis flow meter and the turbine meter

were separated by a flow conditioner.

The data acquisition system included with the TSP was only used to set the test flows

and monitor the TSP; it was not used by the participating NMI to aid perform any aspect

of the calibration. The data acquisition system (data logger and notebook computer)

monitored the transfer standard by collecting flow signals from the flow meters, and

pressure and temperature transmitters in the TSP.

The KC test procedure was designed to enable the quantification of the TSP

performance characteristics [i.e., hysteresis,8 repeatability,

9 reproducibility

10 (turn-on-

8 Hysteresis: maximum difference between the upscale and downscale readings on the same artifact

during a full range traverse in each direction. The standard uncertainty for hysteresis is

3/max downscaleupscalehysterisis YYs −= .

Determining and Reporting Measurement Uncertainties, National Conference of Standards

Laboratories RP-12, ed. 1994.

9 Repeatability (of results of measurements): closeness of the agreement between the results of

successive measurements of the same measurand carried out under the same conditions of

measurement. Def. 3.6, International Vocabulary of Basic and General Terms in Metrology (VIM), ed.

1993.

Figure 1. Design of the Transfer Standard Package.


17 May 2007 6 of 75

turn-off and take-out-put-back)] and their separation from those of the NMI calibration

system. The NMIs were asked to use their routine procedures, personnel, instruments,

software, etc. to calibrate the TSP at the specified test conditions. Details of the test

procedure are provided in the next section.

4 Test Procedure

The test procedure for CCM.FF-K1 was designed to yield laboratory comparison results

having the lowest reasonably attainable level of uncertainty, because at low levels of

uncertainty, the laboratory differences are more clearly quantifiable. This way, not only

will the participants be able to understand and agree with the results of the comparison,

but also the users of the results (i.e., the international traders) will be able to clearly see

and use the conclusions, efficiently and effectively.

To conduct this test in the highly stable way desired, the (TSP) was calibrated according

to the same procedure at each participating laboratory using their routine flow meter

calibration methods. The two meters in tandem provided redundancy that contributed to

the confidence that can be placed on the results obtained. However, flow laboratories do

not typically calibrate flow meters in tandem. Thus, in what follows, emphasis is placed

on the results obtained from upstream meter since those results are more typical of the

data normally obtained in the participating laboratories.

To best compare the participating laboratories, the routine calibration method used at

each laboratory was replicated several times according to statistical recommendations.

The replications enabled the quantification of the TSP’s repeatability9 (including

hysteresis8) and reproducibility

10 under turn-off-turn-on (TOTO) conditions. The

replications also yielded knowledge pertinent to flow meter performance under

changing flow conditions –including cessation of flow– which is pertinent to the use of

these instruments in field applications.

During the KC testing, we thought also important to consider an additional type of

reproducibility that is not typically considered: the take-out-put-back (TOPB)

reproducibility. This type of reproducibility accounts for the change in measurement

10 Reproducibility (of results of measurements): closeness of the agreement between the results of

successive measurements of the same measurand carried out under changed conditions of

measurement. Def. 3.7, International Vocabulary of Basic and General Terms in Metrology (VIM), ed.

1993.


17 May 2007 7 of 75

conditions that result from the removal of the TSP from the pipeline at the end of each

NMI test. In a KC like this one, this type of reproducibility could be significant as the

TSP is installed and removed from the calibration pipeline numerous times, each time

leading to a slightly different measurement system. If not evaluated, this variation,

which is intrinsically contained in the KC data, could then be misinterpreted as NMI-to-

NMI variation when it is really a KC feature.

To quantify the previously mentioned aspects of KC testing, a test procedure was

designed (see Figure 2). In this procedure, each of the two flow meters was tested in

4 groups of 5 measurements each at two different flows, on two different days. As a

result, this two-day test procedure yields 80 measurements from each flow meter. In our

estimation, the four individual sets of 20 measurements (each flow meter tested at the

upstream or downstream positions while operating at high or low flow) were considered

sufficient to adequately estimate the flow meter’s repeatability (including hysteresis),

and the TOTO and TOPB reproducibility. As far as the routine laboratory methods of

testing was concerned (consider only the flow meter in the upstream position),

measurements were obtained using the coriolis flow meter at both flows (1AH, 1AL,

1BL, and 1BH on Day 1) with cessation of flow between them (useful in determining

1 1/A-H&L ON/ H-L/OFF

Statistically Sufficient (Swapped) Multiple-Meter Transfer Standard Tests

DAY Cfg/Tests FLOWS* METERING ARRANGEMENT

2 1 FDS

FLOW

2 1 FDSFLOW

1 2 FDS

FLOW

1 2 FDS

FLOW

1 1/B-L&H ON/ L-H/OFF

1 2/C-L&H ON/ L-H/OFF

1 2/D-H&L ON/ H-L/OFF

2 1 FDS

FLOW

2 1 FDS

FLOW

1 2 FDS

FLOW

1 2 FDS

FLOW

2 2/E-H&L ON/H-L/OFF

2 2/F-L&H ON/L-H/OFF

2 1/G-L&H ON/L-H/OFF

2 1/H-H&L ON/H-L/OFF

Disassemble and Reassemble

+At each flow (Re No.), 5 “FDS runs” are done to produce “factors” for meters #1 & #2;

“H” and “L” refer to the selected hi and lo test flows; these flows are Cardinal test conditions.

Figure 2. Test procedure.


17 May 2007 8 of 75

the TOTO reproducibility). Similar data was also obtained on Day 2 (1GH, 1GL, 1HL,

and 1HH) and used to determining the TOPB reproducibility.

Based on the reported characteristics of the flow facilities of the participating NMIs, it

was anticipated that the TSP would be tested over a wide range of water temperatures.

In fact, several of these NMIs have outdoor facilities that experience temperatures

fluctuations from 10 oC to 35

oC, depending on prevailing weather. One way of

reducing the effects of such large test temperature differences on KC by setting the flow

at the specified Reynolds number,

νVD ⋅

=Re (1)

where V is the mean flow velocity in the TSP, D is the internal diameter of the TSP,

and ν is the kinematic viscosity of the water flowing through the TSP. In (1) we

assume the diameter, D , to expand with temperature, T , according to

( )[ ]00 1 TTDD −+= α (2)

where α is the linear thermal expansion coefficient, and the subscript 0 refers to a

reference condition. In this analysis we selected to neglect the diametric variation of the

TSP with pressure.

In spite of the use of this approach for setting testing flows (also known as cardinal test

points), the temperature fluctuations could affect the measurement principle of the flow

meters themselves and appear as NMI to NMI differences in the results of the KC. To

avoid that, we selected flow meters having measurement principles known to have low

temperature dependency.

At each cardinal test point, the NMIs were asked to determine, using their normal flow

meter calibration procedures, the K -factors for the two flow meters. That is,

VNKV /= (3)

for the turbine meter and,

mNKM /= (4)

for the coriolis meter. In (3) and (4), V and m are the volume and mass of water which

passed through the flow meters, respectively, in the time required for the flow meters to


17 May 2007 9 of 75

generate N pulses (note that N for the turbine meter may be different from N for the

coriolis meter).

The cardinal test points used for this KC were selected to reside in the relatively flat

regions of the performance curves for these flow meters (i.e., portions of the response

curve of the meters where the K -factor does not significantly change as a function of

the Reynolds number). This selection reduced the need to perfectly match the Reynolds

number between laboratories. However, we found during testing that the equilibrated

flow conditions slightly deviated from the target Reynolds number and to correct for

these deviations, the measurements from each NMI were used to produce an estimate of

the slope of the response curve of the meters, thus enabling the projection of the

measurements to the cardinal test points. This method enhanced both the averaged

results for K -factors and reduced the scatter of results due to “off cardinal test points”

tests. Thus adjusted results for both meters, in both positions, are hereafter used.

Coriolis Meter (Error bars are 1 St. Dev)

Run No.

0 5 10 15 20 25 30

KM

12.03

12.04

12.05

12.06

12.07

12.08

uplo

dnlo

uphi

dnhi

uplo avg

dnlo avg

uphi avg

dnhi avg

+/- 0.05 %

Figure 3. Reproducibilities of coriolis meter in 4 flow conditions. Data taken at KRISS.

uplo: meter in the upstream position while operating at low flow conditions;

dnlo: downstream-low flow; uphi: upstream-high flow; dnhi: downstream-

high flow.


17 May 2007 10 of 75

Results obtained at the pilot laboratory for the test produce described above are shown

in Figure 3 and Figure 4. In these figures, the K -factor is shown in the ordinate and the

number of the measurement is shown in the abscissa. The results can be use to obtain

four independent determinations of the repeatability of the flow meters (each using

groups of 5 measurements, i.e., 1-5, 6-10, 11-15, and 16-20), two independent

determinations of the TOTO reproducibility (using measurements 1-10 and 11-20), and

a determination the TOPB reproducibility (using measurements 1-20). Averaged values

for the 20 measurements (with 1 standard deviation error bars) are shown in the right-

hand-side of both figures and ±0.05 % (Figure 3) or ±0.02 % (Figure 4) scatter are

shown.

Figure 3 shows that the repeatability (with hysteresis) of the coriolis meter is relatively

low as compared to its TOTO and TOPB reproducibility. In Figure 4 we can see that the

repeatability and reproducibility of the turbine meter are far better that those obtained

from the coriolis flow meter, which are shown in Figure 3.

Turbine Meter (Error bars are 1 St. Dev)

Run No.

0 5 10 15 20 25 30

KV

10.49

10.50

10.51

10.52

10.53

10.54

uplo

dnlo

uphi

dnhi

uplo avg

dnlo avg

uphi avg

dnhi avg

+/- 0.02 %

Figure 4. Reproducibilities of turbine meter in 4 flow conditions. Data taken at

KRISS. uplo: meter in the upstream position while operating at low

flow conditions; dnlo: downstream-low flow; uphi: upstream-high

flow; dnhi: downstream-high flow.


17 May 2007 11 of 75

5 Measurement Uncertainty

The uncertainties of the K -factors, u ( jK ), ( 1=j for the coriolis meter and 2=j for

the turbine meter respectively), may be expressed as equations (5) and (6).

2/1

2

2

2

2

1 )()()()(

∂∂

+

∂∂

== NuN

Kmu

m

KKuKu MM

M (5)

2/1

2

2

2

2

2 )()()()(

∂∂

+

∂∂

== NuN

KVu

V

KKuKu VVV

(6)

in which m and V are mass and volume of water passing through flow meters

respectively, and N is the number of pulses generated from the meters (not necessarily

the same number for both flow meters). The uncertainty values of K -factors are

obtained from uncertainty budgets of participating NMIs.

During testing, when a single flow meter was calibrated using the flow standard at a

particular NMI, the replications at each test condition enabled calculations of standard

deviations and variances. These standard deviations and variances include contributions

from the “meter” being calibrated and from the standard, or the “system”, and it should

be noted that these two contributions cannot be separated. However, when the two

meters were calibrated simultaneously in tandem – as described above – it is then

feasible to separate the variance contributions from “meter” and “system”, using the

following numerical technique.

After the K -factors from each of the simultaneously calibrated flow meters are

averaged as follows:

N

K

K

N

ij

j

∑= =1

1

(7)

the variances are give by:

( ) 2and,1whichin1

1)(

1

22 =

−−

= ∑=

jKKN

KsN

ijjij (8)

and the correlation coefficient for the simultaneously determined meter factors are given

by:


17 May 2007 12 of 75

( )( )

( ) ( )2/1

1

2

22

1

2

11

1

2211

12

−

−

−−

=

∑∑

∑

==

=

N

i

i

N

i

i

N

i

ii

KKKK

KKKK

r (9)

where the variances of means of K -factors are estimated by,

( ) 2and1whichin20/)(22 == jKsKs jj (10)

Since the coriolis meter and the turbine meter were tested in tandem, the meter factors

of these flow meters are correlated. As the variance, ( )jKs 2 , is sum of the meter

variance and the system variance, then if correlated portion can be attributed to the

common source to both flow meters, i.e., the “system”, then the following is true:

=

jKs

systemj

Ksr 2/22,1

(11)

⋅±=

jKsr

systemj

Ks 22,1

(12)

⋅

−±=

jKsr

meterjKs 2

2,11 (13)

It follows that the combined uncertainty for the meter factor,

jK

cu , of the

measurement is given by,

2/122

+

=

systemj

Kuj

Kuj

Kcu (14)

6 Pilot Comparison Tests

The pilot comparison performed among the pilot laboratory (KRISS-Korea) and pivot

laboratories (CENAM-Mexico and NEL-UK), provided information on the multi-lab

reproducibility of the measurement system (i.e., the TSP and the test procedure). During

these tests, the flow meters were: (i) a coriolis meter manufactured by Endress+Hauser

(model Promass 80FD1600), and (ii) a turbine meter manufactured by Exact Flow

(model EFM64DR-W-C-T1-B-X0 1007).

Any changes in flow meter performance associated with transportation of the TSP

during travel in between laboratories were an important concern to this key comparison


17 May 2007 13 of 75

and were considered as part of the pilot comparison tests. In order to check the change

in flow meter performance due to travel, a final set of tests was performed at the pilot

laboratory after the tests at NEL; hereafter these results are labeled as KRISS-Check.

The mean test temperatures of the initial test performed at KRISS, as well as of the

KRISS-Check tests, were 20.4 oC and 26.5

oC respectively. The average coriolis meter

MK (mass flow rate base K -factor) and the average turbine meter VK (volume flow

rate base K -factor) for 4 different flow conditions for each NMI are shown in Table 2

and Table 3 with their corresponding uncertainties; the same data is illustrated in Figure

5 through Figure 12. In what follows, the medians of the average MK and VK were

chosen as reference values. In the figures, the red dashed lines indicate +/0.07 %

uncertainty limits from the median. It can be seen that the average MK and VK values

of participating NMIs at different flow conditions did not deviate from the median

values by more than +/-0.07 %. Also shown in the data are the test temperatures, which

ranged from 14 oC to 27

oC, but their effects appear not significant in the results.

Table 2. Pilot comparison results, MK and VK , of meters in upstream.

cuphi11 tuphi

12

NMI cuphix , [1/kg] ( )cuphixU , [1/kg] tuphix , [1/L] ( )tuphixU , [1/L]

KRISS 12.0606 0.0064 10.5192 0.0075

CENAM 12.0452 0.0025 10.5125 0.0055

NEL 12.0434 0.0052 10.5176 0.0059

KRISS-Check 12.0601 0.0066 10.5210 0.0075

cuplo13 tuplo

14

cuphox , [1/kg] ( )cuphoxU , [1/kg] tuphox , [1/L] ( )tuphoxU , [1/L]

KRISS 12.0528 0.0039 10.4996 0.0058

CENAM 12.0457 0.0021 10.4939 0.0056

NEL 12.0528 0.0030 10.4944 0.0059

KRISS-Check 12.0551 0.0043 10.5011 0.0062

11 Coriolis meter in upstream at the high cardinal Reynolds number.

12 Turbine meter in upstream at the high cardinal Reynolds number.

13 Coriolis meter in upstream at the low cardinal Reynolds number.

14 Turbine meter in upstream at the low cardinal Reynolds number.


17 May 2007 14 of 75

Table 3. Pilot comparison results, MK and VK , of meters in downstream.

cdnhi15 tdnhi

16

NMI cdnhix , [1/kg] ( )cdnhixU , [1/kg] tdnhix , [1/L] ( )tdnhixU , [1/L]

KRISS 12.0558 0.0064 10.5318 0.0079

CENAM 12.0488 0.0018 10.5228 0.0056

NEL 12.0539 0.0036 10.5252 0.0059

KRISS-Check 12.0616 0.0060 10.5327 0.0075

cdnlo17 tdnlo

18

cdnhox , [1/kg] ( )cdnhoxU , [1/kg] tdnhox , [1/L] ( )tdnhoxU , [1/L]

KRISS 12.0553 0.0040 10.5121 0.0060

CENAM 12.0467 0.0019 10.5045 0.0056

NEL 12.0444 0.0030 10.5006 0.0059

KRISS-Check 12.0575 0.0040 10.5121 0.0063

15 Coriolis meter in downstream at the high cardinal Reynolds number.

16 Turbine meter in downstream at the high cardinal Reynolds number.

17 Coriolis meter in downstream at the low cardinal Reynolds number.

18 Turbine meter in downstream at the low cardinal Reynolds number.


17 May 2007 15 of 75

Figure 5. Averaged values of MK for of the coriolis meter in the upstream position at

high flow.

Figure 6. Averaged values of MK for the coriolis meter in upstream position at low

flow.

Coriolis MeterUpstream @ Low Flow Rate

Test No.

0 1 2 3 4 5

KM

12.04

12.05

12.06

+ 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check

Coriolis MeterUpstream @ High Flow Rate

Test No.

0 1 2 3 4 5

KM

12.035

12.040

12.045

12.050

12.055

12.060

12.065

median

+ 0.07 %

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 16 of 75

Figure 7. Averaged values of VK for the turbine meter in upstream position at high

flow.

Figure 8. Averaged value of VK for the turbine meter in upstream position at low

flow.

Turbine MeterUpstream @ High Flow Rate

Test No.

0 1 2 3 4 5

KV

10.505

10.510

10.515

10.520

10.525

10.530

+ 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check

Turbine MeterUpstream @ Low Flow Rate

Test No.

0 1 2 3 4 5

KV

10.48

10.49

10.50

10.51

+ 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 17 of 75

Figure 9. Averaged values of MK for the coriolis meter in downstream position at

high flow.

Figure 10. Averaged values of MK for the coriolis meter in downstream position at

low flow.

Coriolis MeterDownstream @ High Flow Rate

Test No.

0 1 2 3 4 5

KM

12.04

12.05

12.06

12.07

+ 0.07 %

- 0.07 %

median

KRISS

CENAM

NEL

KRISS-Check

Coriolis MeterDownstream @ Low Flow Rate

Test No.

0 1 2 3 4 5

KM

12.04

12.05

12.06 + 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 18 of 75

Figure 11. Averaged values of VK for the turbine meter in downstream position at

high flow.

Figure 12. Averaged values of VK for the turbine meter in downstream position at low

flow.

Turbine MeterDownstream @ High Flow Rate

Test No.

0 1 2 3 4 5

KV

10.515

10.520

10.525

10.530

10.535

10.540

+ 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check

Turbine MeterDownstream @ Low Flow Rate

Test No.

0 1 2 3 4 5

KV

10.490

10.495

10.500

10.505

10.510

10.515

10.520

+ 0.07 %

median

- 0.07 %

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 19 of 75

The variation in the reproducibility of the coriolis and turbine meter during

the pilot comparison are shown in Figure 13 through Figure 16. These

values (here given in 1-standard deviation) were determined according to

(13). For the turbine meter, the reproducibility did not exceed 0.03 %, but

for the coriolis meter, the values reached about 0.2 % when the meter was in

the upstream position at the high flow. Outliers among the coriolis meter

data at NEL in upstream position at high flow yielded about four times

larger reproducibility than those seen in the other laboratories. But, in the

absence of outliers, the reproducibility of the coriolis meter was less than

0.08 % for all four flow conditions.

Figure 13. Reproducibility variation of flow meters upstream at low flow.

Meters Upstream@ Low Flow Rate

Test No.

0 1 2 3 4 5

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0.08

Coriolis Meter

Turbine Meter

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 20 of 75

Figure 14. Reproducibility variation of flow meters upstream at high flow.

Figure 15. Reproducibility variation of flow meters downstream at low flow.

Meters Downstream@ Low Flow Rate

Test No.

0 1 2 3 4 5

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

Coriolis Meter

Turbine Meter

KRISS

CENAM

NEL KRISS-Check

Meters Upstream@ High Flow Rate

Test No.

0 1 2 3 4 5

Reproducibility

1 Std Dev in %

0.00

0.05

0.10

0.15

0.20

0.25

Coriolis Meter

Turbine Meter

KRISS

CENAM

NEL

KRISS-Check


17 May 2007 21 of 75

Figure 16. Reproducibility variation of flow meters downstream at high flow.

Based on the data above, we felt that the coriolis flow meter (E&H, Promass 80), which

had a reproducibility of about 0.08 %, needed to be replaced by a model of better

reproducibility in order to compare NMIs who claimed uncertainties of better than

0.1 %. We contacted Endress+Hauser Flowtec AG and three new coriolis meters of

improved flow characteristics (Promass 83 F model) were provided. Test results of

these new coriolis flow meters at the Pilot Lab yielded much improved reproducibilities

at the level of about +/-0.05 %.

7 Key Comparison (KC) for Water Flow: CCM.FF-K1

7.1. Description of the TSP for K1

The TSP used in CCM.FF-K1 was similar in design to that used during the pilot

comparison multi-lab reproducibility tests. However, for the better reproducibility, a

new coriolis meter of improved performance (E&H, model Promass 83 F D7) was used

and its flow characteristics are shown in Figure 17. The turbine meter used was an

Exact Flow model 1005 and its flow characteristics are shown in Figure 18.

Meters Downstream@ High Flow Rate

Test No.

0 1 2 3 4 5

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

0.06

0.07

Coriolis Meter

Turbine Meter

KRISS CENAM

NEL

KRISS-Check


17 May 2007 22 of 75

Figure 17. Flow characteristics of the coriolis meter.

Figure 18. Flow characteristics of the turbine meter.

Coriolis Meter (E&H Promass 83F D7)

Reynolds No.

1e+5 2e+5 3e+5 4e+5 5e+5 6e+5 7e+5

KM

11.98

11.99

12.00

12.01

12.02

Experimental values

+/- 0.05 %

Low Cardinal Point

High Cardinal Point

Exact Flow Turbine Meter 1005

Reynolds No.

1e+5 2e+5 3e+5 4e+5 5e+5 6e+5 7e+5

KV

10.28

10.30

10.32

10.34

10.36

10.38

10.40

Experimental values

+/- 0.05 %

Low Cardinal Point

High Cardinal Point


17 May 2007 23 of 75

7.2. Technical Protocol

Based on the test procedure in Figure 2 and the pilot comparison experience, the

technical protocol for CCM.FF-K1 was drafted and modified based on comments from

the participating NMIs before being approved by the CCM.

8 KC Results

8.1. Reynolds Number Correction

The flow meter factors reported by participating NMIs were spread over a range of

Reynolds numbers anchored by the cardinal Reynolds numbers presented in Table 4.

Table 4. Cardinal flow points.

Cardinal Reynolds No. Flow Rate (m3/h)

High Flow 561569 154

Low Flow 254269 70

As shown in Figure 17 and Figure 18, both the coriolis and turbine flow meter have

fairly linear flow characteristics over the range of Reynolds numbers from 250000 to

580000. The reported 40 K -factors from each participating NMI ( MK or VK values of

respective meters at high and low Re numbers in the same configuration) were fitted

linearly in order to relate the reported meter factors to their corresponding cardinal Re

numbers. This procedure was accomplished by linearly fitting the results using the

measured Reynolds numbers and then determining the “effective” K -factor at each

cardinal point using the resulting slopes at each configuration. At each flow condition

(i.e., the same configuration and flow) the average of 20 corrected K -factors, for each

meter, is hereafter presented as the result of the corresponding NMI.

8.2. Results of Participating NMIs

8.2.1. Reproducibility variation

When the TSP returned to the pilot lab it was found to be severely contaminated by fine

rust particles and we decided to clean ultrasonically the turbine meter before the final

check test was performed. However, we decided not to clean the coriolis meter and no

zero point adjustment was needed.


17 May 2007 24 of 75

Figure 19 through Figure 22 show the reproducibilities of the coriolis and turbine flow

meters at each NMI (single lab reproducibility). As seen, there were no significant

changes in reproducibility of the instruments due to contamination of the coriolis meter

or due to the cleaning of the turbine meter. The reproducibility results obtained during

the final tests at the pilot lab for the coriolis meter without cleaning and the turbine

meter after cleaning are labeled in the figures as KRISS-Check.

Figure 19. Reproducibility variation of flow meters upstream at low flow.

Meters Upstream@ Low Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

Coriolis Meter

Turbine Meter

KRISS

sp

PTB

CENAM

NMIJ

NEL

KRISS-Check


17 May 2007 25 of 75

Figure 20. Reproducibility variation of flow meters upstream at high flow.

Figure 21. Reproducibility variation of flow meters downstream at low flow.

Meters Upstream@ High Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

Coriolis Meter

Turbine Meter

KRISS sp

PTB

CENAM

NMIJ

NEL

KRISS-Check

Meters Downstream@ Low Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

Coriolis Meter

Turbine Meter

KRISS

sp

PTB

CENAM

NMIJ NEL

KRISS-Check


17 May 2007 26 of 75

Figure 22. Reproducibility variation of flow meters downstream at high flow.

8.2.2. K -factors

The averaged MK for the coriolis meter and VK for the turbine meter, in upstream

positions at high and low cardinal conditions, are shown in Table 5 and Table 6 for each

of the NMIs with their claimed uncertainties; the average MK and VK values in

downstream positions are presented in Table 4. In these tables, the uncertainties were

estimated using the uncertainty budgets submitted by participating NMIs. As mentioned

before, the data labeled KRISS-Check was obtained during the final test at the pilot

laboratory.

8.2.3. Installation effects

In the tandem flow meter arrangement used in this key comparison, the upstream meter

was always exposed to the typical flow condition seeing at each participating NMI

during a normal calibration. However, because the upstream meter and the flow

conditioner filtered the flow profile entering the downstream meter, its data showed to

be was less influenced by any distortions present in the flow profile of the testing NMIs

(i.e., installation effects). In Figure 23 and Figure 24, the MK values of the coriolis flow

meter in upstream position from each NMI are compared with results for the same flow

Meters Downstream@ High Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

Reproducibility

1 Std Dev in %

0.00

0.01

0.02

0.03

0.04

0.05

Coriolis Meter

Turbine Meter

KRISS

sp

PTB

CENAM

NMIJ

NEL

KRISS-Check


17 May 2007 27 of 75

meter in downstream position (both, at high and low flow). As it can be seen, with the

exception of the data from CENAM, the participating NMIs show very similar MK

values whether the coriolis meter was located in the upstream or downstream position.

This could indicate that the coriolis meter is fairly insensitive to installation effects

influencing its response.

Table 5. Comparison results: MK and VK , of meters in upstream position, 2=k .

cuphi11 tuphi

12

NMI cuphix , [1/kg] ( )cuphixU , [1/kg] tuphix , [1/L] ( )tuphixU , [1/L]

KRISS 12.0121 0.0059 10.3160 0.0074

SP 12.0047 0.0061 10.3047 0.0066

PTB 12.0150 0.0036 10.3099 0.0030

CENAM 12.0040 0.0025 10.3040 0.0070

NMIJ 12.0135 0.0057 10.3083 0.0062

NEL 12.0172 0.0033 10.3126 0.0043

KRISS-Check 12.0179 0.0065 10.3060 0.0078

cuplo13 tuplo

14

NMI cuphox , [1/kg] ( )cuphoxU , [1/kg] tuphox , [1/L] ( )tuphoxU , [1/L]

KRISS 12.0086 0.0038 10.3297 0.0057

SP 12.0039 0.0070 10.3229 0.0054

PTB 12.0116 0.0036 10.3278 0.0030

CENAM 12.0084 0.0027 10.3188 0.0070

NMIJ 12.0153 0.0057 10.3266 0.0062

NEL 12.0122 0.0033 10.3323 0.0046

KRISS-Check 12.0164 0.0041 10.3273 0.0057

However, as indicated by the results shown in Figure 25 and Figure 26, the VK values

of the turbine meter in upstream position are different from those obtained in

downstream position, with the upstream values being lower those obtained in

downstream position. These results lead us to conclude that the turbine meter results

could have been influenced by installation effects during testing.


17 May 2007 28 of 75

Table 6. Comparison results, MK and VK , of meters in downstream position, 2=k .

cdnhi15 tdnhi

16

NMI cdnhix , [1/kg] ( )cdnhixU , [1/kg] tdnhix , [1/L] ( )tdnhixU , [1/L]

KRISS 12.0109 0.0057 10.3253 0.0074

SP 12.0024 0.0060 10.3216 0.0100

PTB 12.0154 0.0035 10.3212 0.0030

CENAM 12.0136 0.0024 10.3169 0.0070

NMIJ 12.0158 0.0058 10.3204 0.0062

NEL 12.0199 0.0058 10.3257 0.0044

KRISS-Check 12.0168 0.0064 10.3113 0.0076

cdnlo17 tdnlo

18

NMI cdnhox , [1/kg] ( )cdnhoxU , [1/kg] tdnhox , [1/L] ( )tdnhoxU , [1/L]

KRISS 12.0078 0.0036 10.3412 0.0060

SP 12.0021 0.0068 10.3392 0.0091

PTB 12.0141 0.0035 10.3398 0.0031

CENAM 12.0156 0.0024 10.3319 0.0070

NMIJ 12.0115 0.0057 10.3369 0.0062

NEL 12.0141 0.0033 10.3441 0.0043

KRISS-Check 12.0135 0.0041 10.3338 0.0060


17 May 2007 29 of 75

Figure 23. MK values of the coriolis meter at high flow.

Figure 24. MK values of the coriolis meter at low flow.

Installation EffectsCoriolis Meter @ High Flow rate

Test No.

0 1 2 3 4 5 6 7 8

KM

11.990

11.995

12.000

12.005

12.010

12.015

12.020

12.025

Upstream position

Downstream position

KRISS

sp

PTB

CENAM

NMIJ

NEL

KRISS-Check

Installation EffectsCoriolis Meter @ Low Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

KM

11.990

11.995

12.000

12.005

12.010

12.015

12.020

Upstream position

Downstream position

KRISS

sp

PTB

CENAM NMIJ

NEL

KRISS-Check


17 May 2007 30 of 75

Figure 25. VK values of the Turbine meter at high flow.

Figure 26. VK values of the Turbine meter at low flow.

Installation EffectsTurbine Meter @ High Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

KV

10.295

10.300

10.305

10.310

10.315

10.320

10.325

10.330

Upstream position

Downstream position

KRISS

sp

PTB

CENAM

NMIJ

NEL

KRISS-Check

Installation EffectsTurbine Meter @ Low Flow Rate

Test No.

0 1 2 3 4 5 6 7 8

KV

10.30

10.31

10.32

10.33

10.34

10.35

10.36

Upstream position

Downstream position

KRISS

sp

PTB

CENAM

NMIJ

NEL

KRISS- Check


17 May 2007 31 of 75

8.2.4. KCRVs and NMI-to-KCRV differences

In what follows, emphasis will be placed on the upstream flow meter results since these

results are more typical of the data produced in each participating NMI.

As recommended by Cox [2, 3] a chi-squared test was performed to ascertain the

consistency of the results in order to determine whether or not the weighted means of

the data could be used in this key comparison. Unfortunately, the consistency test failed,

and as recommended in such eventuality, the medians of the Coriolis meter MK values

and of the turbine meter VK values from the participating NMIs were used in the

determination of the Key Comparison Reference Values (KCRVs). Table 7 shows the

KCRVs, standard uncertainties and the high and low end values of 95 % confidence

intervals for the results of the comparison. These values were estimated by Monte Carlo

simulations according to the procedure B recommended by Cox [2, 3]. In their

estimation, the KRISS-Check results were excluded.

Table 7. KCRVs of MK [1/kg] and VK [1/L] values for meters in the upstream

positions (values derived from medians).

95 % confidence interval

medianx ( )medianxU Hxmedian , Lxmedian ,

cuphi11 12.0123 0.0017 12.0155 12.0090

tuphi12 10.3093 0.0014 10.3120 10.3066

cuplo13 12.0101 0.0010 12.0121 12.0082

tuplo14 10.3270 0.0014 10.3297 10.3242

The relative differences of the test results from the relevant KCRVs have been

estimated (in percentage terms) by Monte Carlo simulation. For this process we used the

test results in Table 5 with their corresponding KCRVs in Table 7. In Table 8 through

Table 13, simulation results are presented with associated standard uncertainties, high

and low end values of 95 % confidence intervals, and En values. Figure 27 through

Figure 38 are graphical presentation of NMI-to-KCRV differences and En values.

In accordance with Cox [2, 3], 0=En designates complete agreement among

participants, while values of 1>En designate them as discrepant at the 5 % level of

significance. In what follows, totalEn is the geometric mean of individual En values in


17 May 2007 32 of 75

different flow conditions describing the degree of equivalence between a NMI and the

KCRV and thus, it shows degree of equivalence between a NMI and the KCRV.

( )( )

( ) 4/141,

2iitotal

i

ii EnEn

du

dAbsEn =∏== (15)

Figure 27. KRISS-to-KCRV differences.

Table 8. KRISS-to-KCRV differences in %.


id ( )idu ),( HdU i ),( LdU i iEn

cuphi11 -0.0018 0.0176 0.038 -0.033 0.05

tuphi12 0.0650 0.0371 0.070 -0.069 0.88

cuplo13 -0.0129 0.0140 0.024 -0.031 0.46

tuplo14 0.0264 0.0267 0.056 -0.045 0.49

totalEn 0.32

KRISS-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Differences

from KCRV

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15

M1/Lo: Coriolis Meter @ Low flow rate, M2/Lo: Turbine Meter @ Low flow rate

M1/Hi: Coriolis Meter @ High flow rate, M2/Hi: Turbine Meter @ High flow rate

M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 33 of 75

Figure 28. En values of KRISS of two flow meters at two different flows.

KRISS-to-KCRV

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

M1/Lo

M2/Lo

M1/Hi

M2/Hi

M1/Lo: Coriolis Meter @ Low flow rate, M2/Lo:Turbine Meter @ Low flow rateM1/Lo: Coriolis Meter @ High flow rate, M2/Lo:Turbine Meter @ High flow rate


17 May 2007 34 of 75

Figure 29. SP-to-KCRV differences.

Table 9. SP-to-KCRV differences in %.



cuphi11 -0.063 0.029 0.056 -0.055 1.10

tuphi12 -0.044 0.032 0.055 -0.063 0.70

cuplo13 -0.052 0.029 0.055 -0.055 0.88

tuplo14 -0.039 0.027 0.047 -0.053 0.73

totalEn 0.84

SP-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

from KCRV

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15



M1/Lo M2/Lo M1/Hi M2/Hi


17 May 2007 35 of 75

Figure 30. En values of SP of two flow meters at two different flows.

SP-to-KCRV

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

M1/Lo M2/Lo

M1/Hi

M2/Hi



17 May 2007 36 of 75

Figure 31. PTB-to-KCRV differences.

Table 10. PTB-to-KCRV differences in %.



cuphi11 0.022 0.018 0.037 -0.032 0.62

tuphi12 0.006 0.015 0.030 -0.029 0.20

cuplo13 0.012 0.013 0.029 -0.023 0.46

tuplo14 0.008 0.015 0.031 -0.030 0.26

totalEn 0.35

PTB-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

from KCRV

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15





17 May 2007 37 of 75

Figure 32. En values of PTB of two flow meters at two different flow.

PTB-to-KCRV

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 38 of 75

Figure 33. CENAM-to-KCRV differences.

Table 11. CENAM-to-KCRV differences in %.



cuphi11 -0.069 0.018 0.035 -0.033 1.97

tuphi12 -0.051 0.034 0.060 -0.066 0.75

cuplo13 -0.015 0.011 0.020 -0.024 0.64

tuplo14 -0.079 0.036 0.072 -0.068 1.10

totalEn 1.01

CENAM-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

from KCRV

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10





17 May 2007 39 of 75

Figure 34. En values of CENAM of two flow meters at two different flows.

CENAM-to-KCRV

Teat No.

0 1 2 3 4 5

En

0.4

0.6

0.8

1.0

1.2

1.4

1.6

1.8

2.0

2.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 40 of 75

Figure 35. NMIJ-to-KCRV differences.

Table 12. NMIJ-to-KCRV differences in %.



cuphi11 0.010 0.020 0.042 -0.035 0.25

tuphi12 -0.010 0.023 0.047 -0.052 0.21

cuplo13 0.043 0.024 0.046 -0.045 0.88

tuplo14 -0.004 0.023 0.049 -0.049 0.08

totalEn 0.25

NMIJ-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

from KCRV

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15





17 May 2007 41 of 75

Figure 36. En values of NMIJ of two flow meters at two different flows.

NMIJ-to-KCRV

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 42 of 75

Figure 37. NEL-to-KCRV differences.

Table 13. NEL-to-KCRV differences in %.



cuphi11 0.041 0.019 0.036 -0.037 1.08

tuphi12 0.032 0.022 0.044 -0.040 0.72

cuplo13 0.017 0.014 0.028 -0.024 0.62

tuplo14 0.052 0.025 0.048 -0.049 1.02

totalEn 0.84

NEL-to-KCRV DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

from KCRV

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15





17 May 2007 43 of 75

Figure 38. En values of NEL of two flow meters at two different flows.

The overall differences between key comparison values of participating NMIs in Table

5 and KCRVs are presented in Figure 39.

NEL-to-KCRV

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo M1/Hi

M2/Hi


17 May 2007 44 of 75

Figure 39. Differences between KC values and KCRVs.

8.2.5. NMI-to-NMI differences

Using the values in Table 3, differences of KC results between NMIs at the same flow

condition were estimated. In order to make the differences relative, relevant KCRVs

were used as scaling factors. These results are illustrated in Figure 39 through Figure 69

and summarized in Table 14 through Table 28 in terms of En values. As mentioned

before, the totalEn is the geometric mean of the individual En values in different flow

conditions and describes the degree of equivalence between NMIs.

NMIs-to-KCRV DifferencesMeters Upstream

Test No.

0 2 4 6

% Difference

from KCRV

-0.2

-0.1

0.0

0.1

KRISS/M1

KRISS/M2

SP/M1

SP/M2

PTB/M1

PTB/M2

CENAM/M1

CENAM/M2

NMIJ/M1

NMIJ/M2

NEL/M1

NEL/M2

Lo Hi

Lo Hi

Lo Hi

Lo Hi

Lo Hi

Lo Hi

M1: Coriolis Meter, M2: Turbine MeterLo: Low Cardinal Reynolds numberHi: High Cardinal Reynolds number


17 May 2007 45 of 75

Figure 40. Differences between KRISS and SP.

Table 14. KRISS/SP differences in %.

ijd ( )ijdU iEn

cuphi11 0.062 0.071 0.87

tuphi12 0.110 0.096 1.14

cuplo13 0.039 0.066 0.59

tuplo14 0.066 0.077 0.86

totalEn 0.84

KRISS/SP DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.1

0.0

0.1

0.2

M1/Lo: Coriolis Meter @ Low flow rate, M2/Lo: Turbine Meter @ Low flow rateM1/Hi: Coriolis Meter @ High flow rate, M2/Hi: Turbine Meter @ High flow rate



17 May 2007 46 of 75

Figure 41. En values of KRISS/SP differences.

KRISS-to-SP

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

M1/Lo

M2/Lo M1/Hi

M2/Hi



17 May 2007 47 of 75

Figure 42. Differences between KRISS and PTB.

Table 15. KRISS/PTB differences in %.

ijd ( )ijdU iEn

cuphi11 -0.024 0.057 0.42

tuphi12 0.059 0.077 0.76

cuplo13 -0.025 0.044 0.57

tuplo14 0.018 0.063 0.29

totalEn 0.48

KRISS/PTB DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo M2/Lo

M1/Hi

M2/Hi


17 May 2007 48 of 75

Figure 43. En values of KRISS/PTB differences.

KRISS-to-PTB

Test No.

0 1 2 3 4 5

En

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 49 of 75

Figure 44. Differences between KRISS and CENAM.

Table 16. KRISS/CENAM differences in %.

ijd ( )ijdU iEn

cuphi11 0.067 0.053 1.27

tuphi12 0.116 0.099 1.18

cuplo13 0.002 0.038 0.04

tuplo14 0.106 0.088 1.20

totalEn 0.53

KRISS/CENAM DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

0.25


M1/Lo

M2/Lo M1/Hi M2/Hi


17 May 2007 50 of 75

Figure 45. En values of KRISS/CENAM differences.

KRISS-to-CENAM

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 51 of 75

Figure 46. Differences between KRISS and NMIJ.

Table 17. KRISS/NMIJ differences in %.

ijd ( )ijdU iEn

cuphi11 -0.012 0.068 0.17

tuphi12 0.075 0.094 0.80

cuplo13 -0.056 0.058 0.97

tuplo14 0.030 0.082 0.37

totalEn 0.47

KRISS/NMIJ DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 52 of 75

Figure 47. En values of KRISS/NMIJ differences.

KRISS-to-NMIJ

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

M1/Lo

M2/Lo M1/Hi

M2/Hi



17 May 2007 53 of 75

Figure 48. Differences between KRISS and NEL.

Table 18. KRISS/NEL differences in %.

ijd ( )ijdU iEn

cuphi11 -0.042 0.055 0.77

tuphi12 0.033 0.083 0.40

cuplo13 -0.030 0.042 0.71

tuplo14 -0.025 0.072 0.35

totalEn 0.53

KRISS/NEL DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo M2/Lo M1/Hi

M2/Hi


17 May 2007 54 of 75

Figure 49. En values of KRISS/NEL differences.

KRISS-to-NEL

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 55 of 75

Figure 50. Differences between SP and PTB.

Table 19. SP/PTB differences in %.

ijd ( )ijdU iEn

cuphi11 -0.086 0.060 1.44

tuphi12 -0.050 0.070 0.72

cuplo13 -0.064 0.066 0.98

tuplo14 -0.047 0.060 0.79

totalEn 0.95

SP/PTB DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10




17 May 2007 56 of 75

Figure 51. En values of SP/PTB differences.

SP-to-PTB

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 57 of 75

Figure 52. Differences between SP and CENAM.

Table 20. SP/CENAM differences in %.

ijd ( )ijdU iEn

cuphi11 0.006 0.056 0.10

tuphi12 0.007 0.093 0.07

cuplo13 -0.037 0.062 0.60

tuplo14 0.040 0.086 0.46

totalEn 0.21

SP/CENAM DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo

M2/Lo M1/Hi

M2/Hi


17 May 2007 58 of 75

Figure 53. En values of SP/CENAM differences.

SP-to-CENAM

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 59 of 75

Figure 54. Differences between SP and NMIJ.

Table 21. SP/NMIJ differences in %.

ijd ( )ijdU iEn

cuphi11 -0.073 0.071 1.04

tuphi12 -0.035 0.088 0.40

cuplo13 -0.095 0.076 1.25

tuplo14 -0.036 0.080 0.45

totalEn 0.69

SP/NMIJ DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10


M1/Lo

M2/Lo

M1/Hi M2/Hi


17 May 2007 60 of 75

Figure 55. En values of SP/NMIJ differences.

SP-to-NMIJ

Test No

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2 M1/Lo

M2/Lo

M1/Hi

M2/Hi



17 May 2007 61 of 75

Figure 56. Differences between SP and NEL.

Table 22. SP/NEL differences in %.

ijd ( )ijdU iEn

cuphi11 -0.104 0.058 1.79

tuphi12 -0.077 0.076 1.01

cuplo13 -0.069 0.065 1.07

tuplo14 -0.091 0.069 1.33

totalEn 1.26

SP/NEL DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10




17 May 2007 62 of 75

Figure 57. En values of SP/NEL differences.

SP-to-NEL

Test No.

0 1 2 3 4 5

En

0.0

0.5

1.0

1.5


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 63 of 75

Figure 58. Differences between PTB and CENAM.

Table 23. PTB/CENAM differences in %.

ijd ( )ijdU iEn

cuphi11 0.092 0.037 2.48

tuphi12 -0.068 0.074 0.92

cuplo13 -0.033 0.037 0.90

tuplo14 -0.040 0.074 0.54

totalEn 1.03

PTB/CENAM DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 64 of 75

Figure 59. En values of PTB/CENAM differences.

PTB-to-CENAM

Test No.

0 1 2 3 4 5

En

0.0

0.5

1.0

1.5

2.0

2.5

3.0


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 65 of 75

Figure 60. Differences between PTB and NMIJ.

Table 24. PTB/NMIJ differences in %.

ijd ( )ijdU iEn

cuphi11 0.012 0.057 0.22

tuphi12 0.016 0.067 0.23

cuplo13 -0.031 0.057 0.54

tuplo14 0.012 0.067 0.17

totalEn 0.26

PTB/NMIJ DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo

M2/Lo M1/Hi M2/Hi


17 May 2007 66 of 75

Figure 61. En values of PTB/NMIJ differences.

PTB-to-NMIJ

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi M2/Hi


17 May 2007 67 of 75

Figure 62. Differences between PTB and NEL.

Table 25. PTB/NEL differences in %.

ijd ( )ijdU iEn

cuphi11 -0.018 0.040 0.46

tuphi12 -0.026 0.050 0.52

cuplo13 -0.005 0.041 0.12

tuplo14 -0.044 0.053 0.82

totalEn 0.39

PTB/NEL DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo

M2/Lo M1/Hi M2/Hi


17 May 2007 68 of 75

Figure 63. En values of PTB/NEL differences.

PTB-to-NEL

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 69 of 75

Figure 64. Differences between CENAM and NMIJ.

Table 26. CENAM/NMIJ differences in %.

ijd ( )ijdU iEn

cuphi11 -0.079 0.053 1.49

tuphi12 -0.042 0.091 0.46

cuplo13 -0.057 0.053 1.09

tuplo14 -0.076 0.091 0.83

totalEn 0.89

CENAM/NMIJ DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10


M1/Lo

M2/Lo

M1/Hi M2/Hi


17 May 2007 70 of 75

Figure 65. En values of CENAM/NMIJ differences.

CENAM-to-NMIJ

Test No.

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4

1.6


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 71 of 75

Figure 66. Differences between CENAM and NEL.

Table 27. CENAM/NEL differences in %.

ijd ( )ijdU iEn

cuphi11 -0.110 0.034 3.20

tuphi12 -0.083 0.079 1.05

cuplo13 -0.032 0.036 0.89

tuplo14 -0.131 0.081 1.61

totalEn 1.48

CENAM/NEL DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.25

-0.20

-0.15

-0.10

-0.05

0.00

0.05

0.10


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 72 of 75

Figure 67. En values of CENAM/NEL differences.

CENAM-to-NEL

Test No.

0 1 2 3 4 5

En

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5


M1/Lo

M2/Lo

M1/Hi

M2/Hi


17 May 2007 73 of 75

Figure 68. Differences between NMIJ and NEL.

Table 28. NMIJ/NEL differences in %.

ijd ( )ijdU iEn

cuphi11 -0.031 0.055 0.56

tuphi12 -0.042 0.073 0.57

cuplo13 0.026 0.056 0.46

tuplo14 -0.055 0.075 0.74

totalEn 0.57

NMIJ/NEL DifferencesMeters Upstream

Test No.

0 1 2 3 4 5

% Difference

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15


M1/Lo

M2/Lo M1/Hi M2/Hi


17 May 2007 74 of 75

Figure 69. En values of NMIJ/NEL differences.

9 Summary and Conclusions

A water flow key comparison, known as CCM.FF-K1, was performed successfully.

From each participating NMI, test results from a Coriolis meter and a turbine meter at

4 different flow conditions were reported with corresponding uncertainty budgets.

Uncertainties of the key comparison results were estimated based on the uncertainty

budgets reported by the participating NMIs. KCRVs and medians from key comparison

results for the Coriolis meter and turbine meter at four different flow conditions were

estimated using Monte Carlo simulation; the final check test results at the Pilot Lab

were excluded in these calculations.

In order to describe the degree of equivalence between key comparison results from

each NMI, a totalEn – describing the overall degree of equivalence between the NMIs

NMIJ-to-NEL

Test No

0 1 2 3 4 5

En

0.0

0.2

0.4

0.6

0.8

1.0

1.2


M1/Lo

M2/Lo

M1/Hi M2/Hi


17 May 2007 75 of 75

and KCRV – was estimated by using the geometric mean of iEn values for two flow

meters at high and low Cardinal test points. totalEn values are listed in Table 29 and are

less than 1.0 for KRISS, SP, PTB, NMIJ, and NEL, while the results of CENAM

yielded a value of 1.01. Nonetheless, the results from this key comparison from all

participating NMIs support their CMC claimed uncertainties, which in all cases are

larger than those reported in this key comparison.

Table 29. Overall degree of equivalence for the participating NMIs

Participating NMI Entotal

KRISS 0.32

SP 0.84

PTB 0.35

CENAM 1.01

NMIJ 0.25

NEL 0.84

10 Acknowledgements

The authors thank K.A. Park and Y. R. Yoon from KRISS; D. Dopheide from PTB; and

R. Paton from NEL who have contributed to this comparison either by analyzing and

discussing the results as well as a critical reading of the manuscript, or by characterizing

the transfer standard and carrying out flow measurement. We also thank M. Anklin

from Endress+Hauser Flowtec Ag for donating three Coriolis meters for the key

comparison. This document was edited by P.I. Espina from the BIPM.

11 References

1 Guidelines for CIPM key comparisons, 1999 with modification by CIPM, in October

2003.

2 M. Cox, “A discussion of approaches for determining a reference value in the

analysis of key-comparison data”, in Advanced Mathematical and Computational

Tools in Metrology IV, Edited by P. Ciarini, A.B. Forbes, F. Pavese, and D. Richter,

pp45-65, 2000, World Scientific Publishing Com.

3 M. G. Cox, “The evaluation of key comparison data”, Metrologia 39, pp 589-595,

2002.

Documents

CCM.FF-K1 for Water Flo · CCM.FF-K1 for Water Flow Final Report 17 May 2007 4 of 75 second change to the schedule occurred when, due to delays in the commissioning of the new water